Written 1999/11/10
Perhaps like me, you might have downloaded a few audio processing programs off the Web and found yourself wishing for less parochial effects. Perhaps you find yourself tiring of reverb, delay and flanger algorithms, and frustrated at the lack of freeware processing programs. Maybe you've hankered to get that Doctor Who, "crappy robot" voice effect on your soundfiles, or heard the terms convolution and spectrum multiplication before but never known what they sounded like. Perhaps you would be interested to know that there is a program that meets all these criteria.
HOG is a freeware audio processing program authored by Brendon Cassidy. His creation was meant to provide the unusual effects often overlooked in many mainstream processing programs. Indeed, for those with a weakness for sonic oddities, the available effects on HOG version 071b beta are a wishlist come to life: convolve, EQ match, ring modulate, frequency extractor, and vocoder. Cool, huh?
The typical setup of the various effects pages has near the top an area to select the input and output files, though the exact configuration will vary depending on which effect you are using. For the output files you can type the filename and choose the directory the file will be saved in. To the right of the filename boxes are buttons labelled with a big yellow letter "P", used to play the corresponding soundfiles. Some effects can make use of 32-bit normalization, and it is highly recommended that you do so as the amplitudes of carelessly processed files can be pretty chaotic. The normalization will keep everything at a nice audible range free of clipping. At the bottom of the average HOG effects page is the process button which you should press when you're ready to hear the output file. There's also a "play when done" option that will play the output file automatically once it has been processed.
The first HOG effect is convolving, less commonly referred to as spectrum multiplication. Now I don't know all the technical details of convolving offhand (you want to go to someone like Electronic Musicians John Deusenberry for that) but the basic gist of it is that that spectral content of an input file of merged with that of an impulse file. The effects can vary widely, considering that the combinations of input and impulse soundfiles are limitless. Because of this you will have to experiment a lot, but once you find certain combinations and settings that work well you can apply them to new input material. I find using the same vocal sound for both input and impulse can be pretty interesting, and if the vocals are done in a ghostly tone to begin with the ouptut file should be remarkably eerie. You may want to take note of the impulse length when you stumble across something striking, as it can have a significant impact on the resulting sound. Alas, I've made some remarkably cool convolved sounds that I cannot reproduce for lack of knowing the exact files and settings I used. Here's a trick I do nowadays to at least make note of the input files: when typing the name for the output file I first type the name (extension excluded) of the input file and then the name of the impulse file, and finish it of with a "con.wav". So if my input is ANTH.WAV and my impulse is 8VCFX.WAV, my output will be named ANTH8VCFXCON.WAV, and I will know the files that were used in creating the output down the road. You might want to take it a step further and include the impulse length with your filename. This naming technique can of course be used for the other effects as well, and will save you heartache should you want to do a variation on a particularly cool effect.
Next is EQ matching, which takes the frequencies of an input and matches them to those of a model file. A human voice EQ matched to an FM synth sound as a model is just one nifty combination. I find that having the number of bands set anywhere from 2048 to 16384 gives you a premium output file, but that's just my own taste.
Ever wanted to know how they made those odd Dalek voices on Doctor Who? Wonder no longer! With the 1*OSCILLATOR setting on HOG's ring modulator, you can recreate that low-budget, British robot voice effect. With this particular setting (my favorite!) you use a basic waveform (sine, sawtooth or square, I prefer sine) to modulate the input file. The frequency of the oscillator is adjustable from 1Hz to 22.05 kHz. Can you say "EXTERMINATE!"? Other modulation configurations include using one file to modulate another and modulating a file by itself offset by an adjustable amount of samples. For these latter setups complex waveforms tend to sound fairly crackled and unappealing. More interesting is throwing a simple synthetic sound or basic waveform into the mix. Because the waveform is simpler it can afford the extra complexity these types of ring modulation add to the output, whereas for the 1*OSCILLATOR type you can process things like speech safely because the modulator is simply a sine, sawtooth, or square. In any event, whether you're perverting vocals or adding extra content to a synthetic sound, HOG does an admirable job indeed.
Decimation is a term for some kind of radical degradation of digital sound, and with HOG, there are numerous ways to decimate. Bit reduction is pretty self explanatory, and creates all sorts of digital noise by lowering the dynamic range. Justify converts the waveform amplitude all to either positive or negative. Try using justify on kick drum sounds. Pre and post-filter let you use either high or lowpass filters with certain cutoff points. Downsample simulates sample rate conversion from .5 to 44.1 kHz, while retaining the original sample rate. That original sample rate part annoyed me, since I'd like to be able to save sample memory with a wave that actually is at a lower sample rate. In this case you'd be better off using some other software to speed up the sound and just playing it at lower pitches. But the HOG downsampling can be useful to audition the effects at least, and one could argue that it eliminates the needs to convert the back to the original rate. All of these decimation effects can be used separately or in combination. For those of you with a taste for the digitally abrasive, this is a must see, I mean, hear.
I must admit the Frequency Extractor is a mystery to me since I don't have any small stereo wav files kickin' around on my hard drive. The basic idea is that it separates the left/right, in/out of phase components of a stereo input file. You can also select model files to include or exclude certain frequencies. That's all I know. You're going to have to try this one for yourself. I have failed you, loyal readers! (sob)
The last effect is the vocoder, which seemed all very promising at first, but I soon found it a little buggy and difficult to get anything typically vocoder-ish out of. The output files never quite sound right to my ear. I've tried numerous combinations of the many parameters and the result was either too human or totally unintelligible. The worst part is that there's almost always a high pitched, cricket-like frequency mixed. It's absent only for the muddiest, most unappealing outputs, but never fails to show up and ruin what would otherwise be half decent (albeit not supremely robotic) results. I suppose you could filter out the unwanted frequency, but I hope future versions make that extra step unnecessary.
In spite of what I thought to be a lacklustre vocoder, and the minor downsampling complaint, HOG is otherwise a great program to twist, tear, and complexify sounds. It does many of the things I had sought, but been unable to find in other freeware sound processing titles in one user-friendly package. Brendon Cassidy has filled an interesting niche with HOG, and he asks nothing in return. What a guy!
You can download a current version of HOG at:
